Digital dentistry is a growing trend with an increasing number of dentists relying on digital impressioning systems. These systems use an intra-oral scanning wand, or scanning of a traditional physical impression, and an associated processing system to generate a digital three-dimensional (3D) model of patients' teeth. In some cases, it may be desirable to compare two digital 3D models of the same patient, acquired with such a system, for diagnostic or other analytical purposes. One of the processing steps in these types of analyses is typically to align or register the two 3D models as closely as possible to one another based on their geometric shapes. The classical approach for registering 3D data is a well-known algorithm called Iterative-Closest Point (ICP) matching, of which there are several variants. However, ICP requires a good initialization to work effectively, meaning the 3D models must already be nearly aligned, and then in most cases ICP merely refines this alignment. In cases where the 3D models are not initially well aligned, or are far apart from one another in position or orientation, ICP often does not adequately work to align or register the two models.
Accordingly, a need exists for registering digital 3D models of teeth for use in displaying the registered models for diagnostic or other purposes.